I'm building financial models for more than 10 years and it is not easy because it takes a lot of time (mostly at night, ....) and it costs a lot of money when you test these models with live data (and your own money)...
Why is it so difficult to move from back testing to live data? Because the model could have been built on specific conditions and/or events (but you don't know it), then it failed when going live.
I'm working on this specific model for more than 3years (with live data), and I try to fine tune it and detect major errors, systematic bad or false signals,...
Recently, I noticed that around 70% of my "bad trades" had been generated by the same parameter. Then, I decided to drop this parameter and to rework the whole model. I had to adapt stop loss/take profit levels and to back test the updated model.
Here are the results of a 2years back testing (all results are in points of the index):
I know it is nothing else than a back testing (at this stage) and this model could fail (once again) with live data.
It is maybe true, maybe untrue... The only way to know it is to test it with my own money....
I will start to run this updated model (live) on Monday the 20th of May 2013, and I will post each morning the updated results on the blog.
Thank you for taking the time to read me, I hope this version is the right one.
For those of you interested in our intraday move, you can visit our site during the day: we post all our trades in real time. You can also subscribe to our twitter account (@SigmaTradingOsc), it is free and you are updated on our latest view/trades.